Eclipse KuDECO

Monday, July 21, 2025 - 20:05 by Rute C. Sofia
This proposal is in the Project Proposal Phase (as defined in the Eclipse Development Process) and is written to declare its intent and scope. We solicit additional participation and input from the community. Please login and add your feedback in the comments section.
Parent Project
Proposal State
Community Review
Background

The evolution of computing from centralized Cloud infrastructures to heterogeneous, decentralised Edge environments has introduced new levels of complexity in orchestrating distributed resources. Traditional orchestration frameworks are often ill-equipped to handle the dynamic, resource-constrained, and context-sensitive nature of edge-cloud systems. As applications demand lower latency, higher resilience, and adaptive deployment strategies, new orchestration paradigms are required.

The Horizon Europe CODECO (Cognitive Decentralised Edge-Cloud Orchestration ) project is a research-driven initiative aimed at enabling intelligent, autonomous orchestration across the IoT-Edge-Cloud (CEI) continuum. The project leverages cognitive computing principles—such as reasoning, learning, and knowledge representation—to provide self-adaptive, policy-aware coordination of services and infrastructure.

Developed within the Eclipse Research Labs, CODECO (codeco-project) incorporates modular components that support automated decision-making for service placement, configuration, monitoring, and adaptation, as a Kubernete framework. It builds on formal models, semantic technologies, and decentralized control strategies to address challenges such as dynamic resource allocation, interoperability, and fault tolerance in edge-cloud environments.

KuDECO reflects the CODECO assets composed of a set of open-source tools to support the deployment of containerized applications across multi-provider federated environments. By proposing KuDECO as an Eclipse project, we aim to foster an open, collaborative platform for researchers, developers, and system architects working on next-generation orchestration solutions. Its integration within the Eclipse ecosystem will promote standardization, reusability, and extensibility, while enabling synergies with complementary projects in cloud management, model-based engineering, and AI for systems.

Scope

Eclipse KuDECO provides an open, extensible platform for the intelligent orchestration of distributed systems across edge and cloud infrastructures. Its primary goal is to enable cognitive, decentralized decision-making for deploying, configuring, and adapting services in highly dynamic and heterogeneous environments.

KuDECO focuses on the following key areas:

  • Cognitive Orchestration
    Decentralised AI techniques that apply reasoning, learning, and knowledge-based inference to automate orchestration tasks, including containerized application placement recommendations to the scheduler, scaling, and reconfiguration/migration.
  • Decentralised Control Models
    Architectures that support distributed decision-making across Edge nodes, Cloud environments, reducing dependency on central control and enabling resilience and autonomy.
  • Semantic Configuration and Policies
    Formal representations (rule based, application-centric) to describe system goals, resource capabilities, context conditions, and deployment policies, enabling adaptive orchestration based on semantic reasoning of applications and resource constrains.
  • Dynamic Context Awareness
    Runtime monitoring and decision support based on context data such as network conditions, resource availability, energy profiles, or user behavior.
  • Modular and Interoperable Architecture
    A component-based design that allows integration with existing orchestration tools, model-driven engineering platforms, monitoring systems, and domain-specific knowledge bases.
Description

Eclipse KuDECO is an open-source Kubernetes-oriented framework designed to enable intelligent, adaptive orchestration of containerised applications across highly dynamic and distributed edge-cloud environments. Unlike traditional cloud-centric orchestrators, Eclipse KuDECO introduces cognitive, decentralised decision-making capabilities that align with the operational demands of modern industrial and cyber-physical systems.

As Edge computing becomes foundational to digital transformation across different competitiveness domains such as Manufacturing, Smart Cities, Eclipse KuDECO addresses the limitations of centralised orchestration platforms like Kubernetes when operating at the network edge, and across an heterogeneous, multi-tenant IoT-Edge-Cloud continuum. KuDECO assumes that  nodes may disconnect, network conditions vary, and latency-sensitive services require rapid, localized decision-making. KuDECO overcomes these challenges by embedding reasoning and context-awareness directly into the orchestration layer.

Key Features:

  • Cognitive Orchestration at the node/cluster Level
    KuDECO augments Kubernetes with decentralized intelligence at each node and at a cluster level, enabling real-time container scheduling based on live context (e.g., CPU usage, network usage, energy consumption, and data freshness).
  • Cross-layer Context Awareness
    KuDECO integrates monitoring of resources, network conditions, and data lifecycle to inform orchestration decisions that meet application-specific goals.
  • Decentralized Architecture
    KuDECO components operate with minimal reliance on central control, promoting scalability, resilience, and autonomy across the edge-cloud continuum.
  • Unified Management via a common operator
    The KuDECO Automated Configuration Management (ACM) component offers a cohesive user interface for developers and cluster managers, managing deployments, configuration policies, and integration with non-Kubernetes systems.
  • Data-Centric Observability
    The Metadata Manager (MDM) provides observability into the full data workflow, treating data as a first-class entity and improving orchestration decisions.
  • Seamless Workload Migration
    While KuDECO can be used with different Kubernetes schedulers, it integrates SWM, which uses a solver-based approach to match containerized workloads with compute, data, and network resources using a min-max graph model.
  • Network Awareness by design
    The Network Management and Adaptability component (NetMA) enables secure, adaptive connectivity and exposes metrics for optimized workload distribution.
  • Privacy-Preserving Learning and Context-Awareness via PDLC
    PDLC is the cognitive “brain” of KuDECO , performing node cost estimation and system stability analysis using decentralized, privacy-preserving learning algorithms.
  • End-to-End Monitoring
    KuDECO offers multi-layered observability, covering system resources (ACM), data workflows (MDM), and network state (NetMA), enabling more intelligent, fault-tolerant orchestration.

Validation and Relevance

KuDECO has been validated in real-world experimental setups, including:

  • Edge clusters using Raspberry Pi and k3s
  • Cloud deployments using scalable virtual machines
  • Test automation via the Horizon Europe CODECO Experimentation Framework (CODEF)

It has demonstrated applicability in several industrial contexts:

  • Real-time orchestration in factory automation
  • Adaptive workloads in smart city scenarios
  • Resilient infrastructure for critical systems

Open Ecosystem

KuDECO is designed for extensibility and collaboration. Its open-source codebase and modular design invite contributions from both academia and industry. Integration with standard Kubernetes environments ensures compatibility and ease of adoption, while its decentralized AI-driven architecture makes it a future-ready alternative for Edge-Cloud orchestration.

Why Here?

The open-source KuDECO framework has already been developed and incubated under the Eclipse Research Labs (codeco-project), where it has served as a platform for research and experimentation in cognitive, decentralized edge-cloud orchestration. As the technology matures and gains traction through successful validation and real-world deployments, transitioning to a formal Eclipse Foundation project under the Eclipse Development Process (EDP) represents the next logical step.

This move enables KuDECO to evolve from a research prototype into a sustainable, production-ready open-source platform, supported by an open governance model and a growing community.

From Research to Open-Source Ecosystem

Establishing Eclipse KuDECO as an official Eclipse project will:

  • Ensure continuity and sustainability of the work initiated in Eclipse Research Labs
  • Provide a clear, open governance structure to guide future development
  • Allow external contributors from academia and industry to engage with confidence
  • Strengthen the project’s alignment with the broader Eclipse ecosystem

Fit with Eclipse’s Strategic Domains

Eclipse KuDECO aligns closely with strategic areas already supported by the Eclipse Foundation, including:

  • Edge and IoT orchestration
  • Cloud-native systems and Kubernetes tooling
  • AI-driven, model-based system management
  • Cyber-physical systems and industrial automation

Its modular architecture and cognitive orchestration capabilities make it complementary to ongoing work across these domains.

Supporting Long-Term Adoption

By formalizing KuDECO as an Eclipse project under the stewardship of fortiss (Coordinator of the Eclipse Research Lab project CODECO):

  • The project gains increased visibility and credibility in open-source and industry circles
  • All code and IP contributions will be managed under the Foundation’s trusted legal and compliance processes
  • The Eclipse community can help shape the direction of the platform, driving future enhancements and integrations

In summary, transitioning CODECO from a research lab initiative to a formal Eclipse project is a natural and strategic evolution. It ensures that the results of high-impact research can continue to grow within an open, collaborative, and production-ready ecosystem.

Future Work

As Eclipse KuDECO transitions from the Eclipse Research project CODECO into a fully open-source, production-grade orchestration platform, several areas of ongoing and planned development will guide its roadmap:

1. Advanced Learning and Adaptation Capabilities

Future iterations of KuDECO aim to strengthen the use of decentralised AI and the integration of context-awareness considering:

  • Enhancing learning algorithms for improved decision quality in dynamic environments
  • Supporting federated learning scenarios for distributed intelligence across edge nodes
  • Incorporating security,  energy-aware QoS and user-defined metrics into the orchestration logic

2. Integration with Eclipse Ecosystem Tools

The project will explore integration opportunities with complementary Eclipse projects, such as:

  • Eclipse EMF and modeling tools for declarative orchestration policies
  • Eclipse Kura for edge device management
  • Eclipse ioFog or similar projects for edge container runtime support

3. Expanded Monitoring and Observability Stack

KuDECO monitoring components (ACM, MDM, NetMA) will evolve to:

  • Provide deeper integration and observability across the application-data-network stack
  • Include pluggable observability pipelines with standard metrics (e.g., OpenMetrics, OpenTelemetry)
  • Enable anomaly detection and predictive alerts based on orchestration KPIs

4. Multi-Cluster and Cross-Domain Orchestration

Enhancements will focus on supporting:

  • Seamless orchestration across multiple Kubernetes clusters (e.g., edge federation)
  • Cross-domain orchestration involving heterogeneous cloud/edge providers and administrative zones
  • Dynamic service discovery and mesh networking between geographically distributed nodes

5. Experimentation Framework and Use Case Expansion

The CODECO Experimentation Framework (CODEF) will be extended to support:

  • Large-scale scenario simulations with synthetic and real-world workloads
  • Integration of new domain-specific use cases (e.g., 5G, critical infrastructure, autonomous vehicles)
  • Benchmarks for evaluating orchestration decisions, energy consumption, and QoS trade-offs

6. Industrialization and Community Growth

Future work will also focus on:

  • Engaging industrial partners to validate KuDECO in production settings
  • Growing the contributor base via documentation, tutorials, and developer guides
  • Establishing a governance model and release process aligned with Eclipse best practices
Project Scheduling

The CODECO framework has already undergone multiple development and release cycles under the Eclipse Research Labs, with functioning prototypes, active components (e.g., ACM, PDLC, NetMA), and real-world validation. 

The transition to a formal Eclipse Foundation project (Eclipse KuDECO) will focus on production-readiness, open governance, and community engagement.

The following phases reflect that maturity while aligning with the Eclipse Development Process (EDP):

Phase 0 – Pre-Provisioning (Completed)

  • Initial development and validation under Eclipse Research Labs
  • Open-source releases on Eclipse GitLab
  • Demonstrators and experimentation frameworks tested on edge/cloud setups
  • Community interest and early stakeholder feedback gathered

Phase 1 – Project Creation and Transition (Months 1–2)

  • Submit and approve the Eclipse KuDECO project proposal under the EDP
  • Formalize project metadata, scope, license (Apache 2.0), and contributor agreements
  • Migrate repository into official Eclipse project structure (if needed)
  • Set up project pages, governance documentation, and CI/CD integration under Eclipse standards

Phase 2 – Stabilization and Community Onboarding (Months 3–4)

  • Review and refactor components for coding standards, modularization, and documentation
  • Prepare first official Eclipse release (e.g., Eclipse KuDECO 1.0) based on already-published research lab artifacts
  • Provide contributor onboarding guides, development workflows, and public roadmap
  • Establish initial project team roles (Committers, component Lead, etc.)

 Phase 3 – Feature Hardening and Ecosystem Integration (Months 5–8)

  • Harden existing components (e.g., PDLC inference engine, NetMA metrics, SWM solver)
  • Enhance monitoring/observability integration (Prometheus, OpenMetrics)
  • Begin integration efforts with relevant Eclipse technologies (e.g., Eclipse Kura, ioFog, EMF)
  • Release updated use cases and deployment profiles across edge-cloud setups

 Phase 4 – Community Growth and Iterative Releases (Months 9–12)

  • Formalize the release cycle (e.g., 6-month cadence) with version tagging and API guarantees
  • Expand community via webinars, demos, and academic/industrial partnerships
  • Conitnue the deployment on diverse environments (k3s, lightweight edge devices, multi-cluster)
  • Explore graduation roadmap or cross-project collaboration within the Eclipse ecosystem
Committers
Pepi Paraskevoulakou (This committer does not have an Eclipse Account)
Dean Kelly (This committer does not have an Eclipse Account)
Alka Nixon (This committer does not have an Eclipse Account)
Luis M. Contreras (This committer does not have an Eclipse Account)
ALEJANDRO TJAARDA DE COCK BUNING SANGUINO (This committer does not have an Eclipse Account)
Interested Parties

The Eclipse KuDECO project benefits from a diverse and well-established consortium of institutions and organizations who have already contributed to its development under the Horizon Europe initiative CODECO:

🏢 fortiss GmbH (Lead & Coordinator, Germany)

Primary technical steward for CODECO and for KuDECO leading the architectural framework design, components development, and experimentation infrastructure.

🏢 Eclipse Foundation Europe GmbH (Germany)

Contributor to open-source strategy, tooling, dissemination campaigns, and Eclipse integration.

🏢 INOVA+ – Innovation Services, SA (Portugal)

SME partner contributing to the Research and Development programme of CODECO (IRCEP) and to its broad dissemination.

🏢 Atos Spain SA & Atos IT Solutions and Services Iberia SL (Spain)

Supporting industrial integration, transfer to standardisation, and the monitoring aspects in CODECO.

🏢 Intracom ICT (Greece)

Developer of the GNN-based infrastructure inference aspects, and leading the CODECO system integration.

🏛️ ATHINA Research & Innovation Center (Greece)

Research partner specializing in advanced networking observability and AI models, and main responsible for the asset CODEF.

🏛️ University of Göttingen (Germany)

Use-case leader (Smart Cities)

🏢 Siemens AG (Germany)

Leader of the open-source component SWM

🏢 Netcompany S.A. (Luxembourg)

Collaborating on decentralized scheduling and federated learning components.

🏞️ Fundació Privada i2CAT (Spain)

Collaborating and leading the decentralised AI aspects as well as a use-case (Mobility).

🏛️ University of Piraeus Research Center (Greece)

Collaborating and leading the decentralised AI aspects; leader of the CODECO Synthetic Data Generator.

🏢 Telefónica Innovación Digital SL (Spain)

Leading the component NetMA and use-case (Smart Cities, CDN)

🏛️ Universidad Carlos III de Madrid (Spain)

Affiliated academic contributor to the NEtMA component

🏛️ Universidad Politécnica de Madrid (Spain)

Use-case leader (Energy)

🏢 Almende BV (Netherlands)

SME partner working on the integration of CODECO to a use-case (Energy)

🏢 Red Hat Israel Ltd & associated Red Hat entities (Israel, Sweden, Ireland, Spain)

Contributing Kubernetes and container orchestration expertise; leading the development of the ACM component.

🏛️ IBM Research GmbH (Switzerland)

Partner institution providing the open-source MDM component.

Initial Contribution

The initial contribution for the Eclipse KuDECO project will consist of a mature codebase and supporting assets developed within the Eclipse Research Labs by the contributors. This contribution has already been released under the Apache License 2.0 and validated through multiple internal releases, deployments, and academic publications.

The initial contribution includes:

Core Framework Components

  • ACM (Automated Configuration Management):
    Unified user interface and control layer for deployment management, user role handling, and integration with Kubernetes clusters.
  • MDM (Metadata Manager):
    Data observability component that tracks data as a first-class orchestration entity considering data sources.
  • SWM (Seamless Workload Migration):
    Solver-based workload scheduler that matches application requirements with compute, data, and network resources using a graph-based approach.
  • NetMA (Network Management Agent):
    Provides secure, SDN-enabled pod connectivity and exposes metrics for network-aware workload placement.
  • PDLC (Privacy-preserving Decentralised Learning and Context-awareness):
    The cognitive core of KuDECO , performing node cost estimation and system stability inference via decentralized AI mechanisms.

Supporting Infrastructure

  • CODECO Experimentation Framework (CODEF):
    A toolset to simulate, deploy, and validate orchestration scenarios at scale using k3s clusters, VMs, and edge-node testbeds.
  • Monitoring stack:
    Integrated with Prometheus and future extensions (e.g., OpenMetrics) for full-stack observability.

Artifacts and Documentation

  • Existing YAML-based KuDECO Application Model
  • Deployment scripts, Helm charts, and Kubernetes manifests
  • Technical documentation and usage examples
  • Initial set of test cases and scenarios used in prior research validations

Development Language and Tools

  • Core implementation in Go (Golang)
  • Container orchestration built on Kubernetes and compatible with standard K8s APIs
  • Use of standard tooling (Helm, Prometheus, k3s) for deployments

Repository Location

The initial contribution is currently hosted at:
🔗 https://gitlab.eclipse.org/eclipse-research-labs/codeco-project

Upon project provisioning, this codebase will be migrated or restructured (if necessary) under the official Eclipse project namespace, and the CI/CD pipeline will be integrated with Eclipse’s build infrastructure.

Source Repository Type